PassiveAggressiveClassifier Fake News Detector
Model Description
This model uses a PassiveAggressiveClassifier from scikit-learn to classify news articles as "Fake" or "Real". The input data consists of news articles from two datasets (True.csv and Fake.csv). Text data is preprocessed (lowercased, punctuation and numbers removed, extra spaces cleaned) and vectorized using TF-IDF. The model is trained on 80% of the data and tested on the remaining 20%.
Intended Uses & Limitations
- Intended use: Detect fake news in English news articles.
- Limitations: May not generalize to other languages or domains. The model is trained on a specific dataset and may not perform well on out-of-distribution samples.
Training Data
- True.csv: Real news articles.
- Fake.csv: Fake news articles.
- Both datasets contain columns:
title, text, subject, date.
Preprocessing
- Lowercasing text
- Removing punctuation and numbers
- Removing extra spaces
- TF-IDF vectorization with English stopwords,
max_df=0.7
Metrics
- Accuracy: 99.41% on the test set.
